Impedements to Building Survivable Systems: An Experience Report
Citation P. Rubel, P. Pal, F. Webber, M. Atighetchi, and C. Jones. IEEE ISW - 2001/2002 workshop.
Formats: PDF
Abstract In the distributed systems technology group at BBN, we have been researching issues involving survivability for the past couple of years. One approach to survivability is to focus on the total prevention of malicious behavior. This results in systems with a closed architecture, where many of the components have to be custom built. This approach is very expensive and most modern useful systems need to be open i.e. need to interact with other systems and their environment. We assume that solutions applicable to open systems are of greater value and focused our research on open systems. We believe that it is possible to make systems more survivable using adaptation that involves interaction with the environment and security mechanisms even though the defense architecture is open.
In this paper we will briefly outline some conceptual impediments to survivability as well as some practical impediments to realizing survivable systems that we have encountered in the course of our work. The conceptual impediments point to deeper research issues in the emerging area of survivability, such as how we define and validate survivability requirements and how we compare and evaluate the survivability characteristics of similar systems. In addition to the issues in mechanisms for achieving survivability there are also a number of practical impediments discussed are well known software engineering issues dealing with developing large software systems. These include getting software components to work together and then how to reuse those components in other systems. We claim that in the context of survivability these issues become even more complicated.
